Number of Parents

156

Mean Age of the Parents

44.76

Percent Male

46.79

Mean Age of Children

11.81

Percent Boys

54.48

Table 1: Number of Children per School and Grade

Level	Grade	N
Primary	6	30
Primary	7	29
Primary	8	26
Secondary	1	29
Secondary	2	20
Secondary	3	22

Introduction

Perceived Appropriateness of Active and Passive Parental Consent in Social Science Research in the School Context



Thabo van Woudenberg, Esther Rozendaal & Moniek Buijzen
Erasmus University Rotterdam

Abstract

The majority of studies on children in social science use active parental consent procedures. This has led to lower-powered studies with lower external validity compared to passive parental consent procedures. Moreover, the total burden, time, and cost are higher in active consent procedures. In this study, we investigated the perspective of parents. Specifically, we used an online survey in which parents were exposed to multiple types of research in video vignettes and were asked to indicate how appropriate they rate active – and passive parental consent. We used a reversed inferior hypothesis to test whether active consent was not rated as more appropriate than passive consent. Results indicated that parents perceived active consent procedures as appropriate in all types of research. Passive consent was rated comparably appropriate for some types of research, particularly for secondary school children. Furthermore, we explored the entire procedure of providing consent, reported in an online dashboard. We conclude by providing recommendations for when and how to use either active or passive parental consent for social science research in the school context.

Sample
Data were collected between the 7th and 12th of April 2022. Recruitment and compensation of the participants were managed by the panel agency PanelClix and were redirected to an online survey in Qualtrics. The initial 211 participants received detailed information about the study and data collection, and consented to voluntary participation in the study. After screen for participants that had no children within the targeted age (between the ages of 8 and 16 years old), speeding and failing the instructed response question, the analytical sample contained 156 parents. For details on the sample see the panel on the left.

Links:
Preprint
Article
Rreregistration
Data
Example vignette

Vignettes

On this dashboard, we visually present the all the findings of our study. We hope that providing the information in this way, we can better inform researchers and ethical review boards to make an informed decision in the study procedures. The main question of this study was how appropriate parents of children in primary- and secondary school perceive active- and passive informed consent for participation of their child in scientific research in the school context. Specifically, this study investigated the difference in perceived appropriateness between active- and passive parental informed consent for a variety of research types conducted in the school context.

To test the main hypothesis, whether passive parental consent is not perceived as less appropriate than active parental consent, we used a reversed inferior test. We defined a one-sided t-test with the alternative hypothesis that active parental consent is perceived as more appropriate than passive parental consent. However, to confirm our hypothesis we were interested in the amount of evidence for the null hypothesis, meaning we would accept our hypothesis when the Bayes factor for the alternative hypothesis is smaller than 3 (BF10 < .33; BF01 > 3).

Participants were exposed to 10 video vignettes, each explaining one of the research types. Participants indicated how appropriate both active and passive parental informed consent is for each vignette. Participants could rate the appropriateness on a 7-point slider scale, ranging from not appropriate at all (1) to very appropriate (7). The ratings of active and passive consent were asked on two separate pages, and each time the illustration of the vignette was presented alongside the research method, see the Figure 1. Table 2 and figure 3 present the mean ratings of passive and active consent, for the ten research types. In addition, the table reports the Bayes factors for the alternative hypotheses and the conclusions that we draw.

Table 2: Perceived Appropriateness of Active and Passive Parental Consent

	Passive		Active
Research type	Mean	95% CI	Mean	95% CI	Bayes factor	Conclusion	Cohens D
Observational	5.06	[4.77, 5.36]	4.74	[4.45, 5.02]	0.04	Pas = Act	0.11
Co-creation	4.69	[4.41, 4.96]	5.29	[5.05, 5.53]	0.40	Pas = Act	0.11
Survey	4.13	[3.82, 4.45]	5.51	[5.26, 5.76]	10.44	Pas < Act	0.23
Longitudinal Survey	3.58	[3.27, 3.89]	5.42	[5.17, 5.68]	27.40	Pas < Act	0.26
Focus group	4.84	[4.55, 5.13]	5.13	[4.88, 5.39]	>100	Pas < Act	0.34
Interview	4.48	[4.18, 4.78]	5.16	[4.91, 5.41]	>100	Pas < Act	0.49
Objective Measures	3.48	[3.17, 3.79]	5.50	[5.25, 5.75]	>100	Pas < Act	0.66
Diary	4.50	[4.20, 4.80]	5.42	[5.18, 5.66]	>100	Pas < Act	0.71
Data Donation	3.54	[3.24, 3.85]	5.46	[5.21, 5.70]	>100	Pas < Act	0.73
Clinical	2.72	[2.41, 3.04]	5.51	[5.20, 5.81]	>100	Pas < Act	0.92

Figure 1: Measurement of the Outcome Variables

Figure 2: Perceived Appropriateness of Active and Passive Parental Consent

Figure 3: Perceived Appropriateness for Primary School Children

Differences Between Primary and Secondary Schools

Because we were interested in potential differences between primary school and secondary school children, we stratified our analyses by the school level. A Bayesian mixed-effects model indicated strong evidence that there is no difference in scores in general between primary and secondary school (BF10 = .08 ± 1.02%). This means that overall (irrespective of the consent procedure or research type), parents rated the acceptability similarly. However, we found anecdotal evidence (BF10 = 2.88 ± 0.93%) that the differences between active and passive consent were bigger in primary school (active = 5.54, passive = 3.91, difference = 1.54) than secondary school (active = 5.15, passive = 4.34, difference = 0.81).

Splitting the vignettes into primary and secondary schools provided the most interesting insights. Specifically, the same Bayesian paired samples t-tests from the preregistered analyses were run for the primary and secondary schools separately. For the primary school children, no different conclusions were drawn. For children in secondary school (n = 71), we found anecdotal to moderate evidence that parents do not think that active consent is more appropriate for Focus groups (BF01 = 2.63 ± 0.00%), Survey (BF01 = 2.08 ± 0.00%), and Longitudinal Survey (BF01 = 3.01 ± 0.00%) studies. This differences is also observable when comparing Figure 3 to figure 4.

Based on these results we can tentatively state that for secondary schools, passive parental consent is an appropriate procedure for Observational, Co-creation, Survey, Longitudinal survey, and Focus group studies.

Figure 4: Perceived Appropriateness for Secondary School Children

How Appropriate is Passive Consent when Participants are Asked About …

Were were interested whether parents would regard certain topics of interested to our projects as more or less appropriate for passive consent. Therefore we asked the parents how appropriate passive consent is for surveys when we ask questions pertaining to their happiness and mental well-being, about their friends, about the situation at home, and personally identifiable information such as address or date of birth. In Table 3 and Figure 5, we have also added the general appropriateness of passive consent from the vignettes. The general label refers to the appropriateness of passive consent for surveys form the vignettes. As indicated in the table and the figure, the scores for happiness and friends were similar to the general sentiment of using passive consent in survey studies. However, parents were a bit more reserved about questions on the domestic situation and especially about personally identifiable information.

Table 3: Specific Themes in Survey and Passive Consent

	Mean	SD	SE	95% CI
Happiness	4.72	1.71	0.14	[4.46, 4.99]
Friend	4.65	1.68	0.13	[4.39, 4.92]
Domestic	4.17	2.01	0.16	[3.85, 4.48]
Personal	3.60	2.12	0.17	[3.26, 3.93]
General	4.69	1.76	0.14	[4.41, 4.96]

Figure 5: Specific Themes in Survey and Passive Consent

Creating Media Content

In our lab, we pride ourselves for using a blended science approach. In our projects, we cooperate with youth to design our studies. Specifically, together with the children we co-create media content such as Instagram posts or TikTok videos. Therefore, we asked the parents how appropriate passive consent is for co-creation when participants will be making photo’s or video’s.
With a mean score of 3.57 this was not perceived as appropriate by the parents, as it is below the mid-point of the scale and considerable lower than the appropriateness of passive consent for co-creation in general (4.69). Therefore, we advise researchers to use active parental consent procedures when co-creating media messages. This is also in line with the previous reservation about personally identifiable information. Most media content could contain some form of information that is identifiable, such as names, places or faces.

Cluster Consent in General

The parents in the study were shown a video about a specific format for obtaining parental consent for multiple studies within one school year. Specifically, parents are asked to provide active parental consent at the start of the school year for a series of studies on a determined topic. For each study within the school year, parents are informed and have the opportunity to opt-out. We have termed this cluster consent. The parents in this study were asked to respond to the idea of the cluster consent in general. Again they used the faces scale and could select the face that best matched their beliefs. The range was between 1 and 7 as shown in Figure 1.
Compared of the average scores of the 10 vignettes, the cluster consent scored a bit higher. But, be aware, these two scores are not entirely comparable. The passive and active scores are the averages of the vignette. The cluster score was just one single item measuring appropriateness in general.

Table 4 : Cluster Consent Compared to Active and Passive consent

	Mean	SD	SE	95% CI
Active	5.31	1.66	0.13	[5.05, 5.57]
Passive	4.10	2.04	0.16	[3.78, 4.42]
Cluster	5.45	1.79	0.14	[5.17, 5.73]

Cluster Consent per Research Type

In the next question, participants gave a bit more information about the cluster consent for the different research methods. Contrary to the vignettes, they did not rate the appropriateness on a scale from 1 to 7, but they could indicate per research type if they thought the cluster consent would be appropriate (yes vs no). Here are the percents of “yes” per research method. As can be seen in Figure 6, the majority of the parents was open to the concept of cluster consent for Survey, Observation, Longitudinal survey, Co-creation, Interviews and Diary studies. Much to our surprise, using cluster consent procedures for Focus group studies was only accepted by the minority of the parents.

Figure 6: Is cluster Consent Appropriate per Research Method

Responding to Invites for Academic Research

Participants were asked whether they have been approached previously to provide consent for their child in academic research. Follow-up questions asked about research why people did not respond previously.

As can be seen in Table 5, the majority of the parents in our sample was never approached before to provide consent for their child to participate in scientific research.

Table 6:

Previously approached	Active	Passive	Difference
No	5.28	4.03	1.25
Yes	5.50	4.22	1.28

But When Approached Before

Of the 37 participants that were approached previously, almost everybody did respond! Only two people answered that they did not respond to a previous invite, see Table 7. In theses cases, participants were provided a text box in which they could explain why they did not respond when approached.

The first participants indicated that (s)he had forgotten to respond. The second participants said (s)he did not respond because a passive consent procedure was used. Based on only these two participants we can not learn a lot. But at least, out of those 37 parents, there were no critical opponents of the consent procedure that they study used.

Table 5:

Previously approached	Number of Participants	Percent
Yes	37	0.24
No	106	0.68
Not sure	13	0.08

The Differences Between Those Who Were Approached Before

Given that so few of our participants were approached before, we wondered if this had an effect on the ratings for active and passive consent. We deemed it plausible that the previous invitation might increased the trust in scientific research, or might have activated more critical thinking.
The ratings of both active and passive consent were slightly higher for those who have been approached previously, see Table 6. However, it is very unlikely that that there is a difference in ratings between parents who have been approached before and parents who haven’t (BF = 0.21). Also, we find no differences specifically for active (BF = 0.28) and passive (BF = 0.21) consent, and the differences between active and passive consent were comparable between both groups (BF = 0.15). Therefore, we can not conclude that having previously been invited to provide for consent your child does not affect the rating for appropriateness for either active of passive parental consent procedures.

Table 7:

Reponded	Number of Participants	Percent
Yes	35	94.59
No	2	5.41

Communicating With Parents

Participants were asked via which channels they like to opt-in or opt-out of the study. Participants could select multiple channels. In the figure, we have plotted the percentage of participants that selected the channel. we have annotated the exact number of participants next the the end of the bar. Because participants could select multiple channels, the total of the percentages does not add up to 100%.
As can be seen in Figure 7, Email and website were the most selected channels. When participants selected other, they could give a suggestion. One of these suggestions was via an app, the other text field was left blank.

Moreover, participants were asked how important it is that communication about the study is done in the same way the school normally communicates with the parents. The participants could respond with not at all (1) to very important (7). On average, the score was 5.82 (SD = 1.28), meaning that most of them agreed that it is important to use the schools’ existing channels of communication, see Figure 8.
In short, we advise to use the schools’ existing channels, preferentially email or website, for the consent procedure.

Firgure 8: Using School’s Communication Channels

Figure 7: How to Communicate With Parents

Sharing the Research Data

Participants were asked whether they have problems with sharing the research data, given that identifiable information is deleted. Specifically, we wanted to know parents view sharing of the data with other colleagues, alongside the academic report or paper, with the other participants of the study or sharing the data online for everybody to see.

Participants could respond the the statements, e.g., I think it is problematic when data is shared with the other participants in a study on a 7 point Likert scale, ranging from completely disagree (1) to completely agree (7). Here, we reflected the values, so that higher scores mean less problematic.

It seems that parents do not think that sharing the data with academic colleagues is problematic. The mean score is above the midpoint on the scale and there is a clear peak at 6. However, sharing the data together with the report, with the other participants of the study or putting the data on a website is seen as a problem by most parents.

Table 8: Sharing the Research Data

	Mean	SD	SE	95% CI
Colleagues	4.34	1.95	0.16	[4.03, 4.65]
Report	3.81	1.86	0.15	[3.52, 4.10]
Participants	3.28	1.87	0.15	[2.99, 3.58]
Open	3.16	1.79	0.14	[2.88, 3.44]

Open Science

This suggest that parents see some merit in sharing data with colleagues within academia, but having the data publicly available is one step too far. So one the one hand this means some support for open science practices. But on the other hand, researchers should limit the use of the data of children only to academia, or explain to the parents of the study why it is important to openly share the de-identified data online. And although Creative Commons does not recommend this, the data could also be shared with a NonCommercial (NC) license to limit commercial use or or a NoDerivatives (ND) lisence so that the data can not be modified or adapted in any way. Downside is that other researchers could only verify the original research, but could not reuse the data tot test new hypotheses (with asking permission to the author).

Figure 9:

Figure 10: Rewarding Participants

Rewarding Children

Participants were asked what they think would be a fair reward for participation in scientific studies. Participants could select select multiple rewards. In the figure, we have plotted the percentage of participants that selected the channel. we have annotated the exact number of participants next the the end of the bar. Because participants could select multiple rewards, the total of the percentages does not add up to 100%.
As can be seen in Figure 10, a gift for the whole class and a gift for the individual child were the most selected rewards. Around one in three participants indicated that it was also not necessary to reward the participants, as it is important to participate in scientific studies. When participants selected other, they could give a suggestion.

One of these suggestions was to donate money to charity chosen by the children. Another suggestion was to handout discount coupons for excursion. The last interesting suggestion was to let the children decide what they want as a reward.

Table 9: Monitary Rewards for Children

Payment	N	Mean	Median	Mode	Min	Max
Vouchers	20	8.12	5	5	1	50
Wire money to child	15	6.17	5	5	1	15
Wire money to parent	2	10.00	10	NA	5	15
Pay in cash	2	5.50	6	NA	1	10

Monetary rewards

In addition, participants who selected some form of momentary reward were asked to add what they thought would be a fair amount of euro’s per hour. As can be seen, the amount vary heavily, between 1€ and 50€ (!) per hour. But looking at the mean (7,33€),median and the mode (both 5€), the results suggests that 5€ per hour is reasonable for the parents.

Thabo van Woudenberg

Thabo van Woudenberg (PhD) is a postdoc researcher at the Erasmus School of Social and Behavioural Science at the Erasmus University Rotterdam. He is a member and scientific coordinator of the SocialMovez project, as part of the Movez Lab. The overall aim of this project is to develop and test a framework for effective and responsible health campaigns, using online social networks to identify and motivate peer influencers while safeguarding digital privacy. His studies in particular focus on social network influences on health-related behaviors in adolescents.

Link to Twitter
Link to LinkedIn
Link to Profile
Link to Personal website

Esther Rozendaal

Esther Rozendaal (PhD) is associate professor of Digital Resilience in the Erasmus School of Social and Behavioural Science at the Erasmus University Rotterdam. She is a member of the Movez Lab, a research team with a shared interest in young people, digital media, and wellbeing. In 2020, she was awarded a Vidi grant from the Dutch Research Council (NWO) for a research project investigating how teenagers can be empowered to use digital media in a safe and responsible manner. Rozendaal is also interested in studying the role of advertising literacy in young people’s responses to advertising.

Link to LinkedIn
Link to Profile

Moniek Buijzen

Moniek Buijzen (PhD) is professor of Communication and Behavioural Change in the Erasmus School of Social and Behavioural Sciences at the Erasmus University Rotterdam and the Behavioural Science Institute at Radboud University, both in the Netherlands. In her research she investigates how to harness the potential of digital media technology to improve young people’s well-being, while minimizing potential risks. Her work has been funded by prestigious grants of the European Research Council (ERC) and the Dutch Research Council (NWO) and has been recognized with awards of several international communication associations. Buijzen strives for a continuous interaction between research and innovative technological applications and embraces a ‘blended science’ approach, integrating research, teaching, and science-society collaboration.

Link to LinkedIn
Link to Profile

Erasmus University

Erasmus University Rotterdam (EUR) is an internationally oriented university with a strong social orientation in its education and research, as expressed in our mission ‘Creating positive societal impact’. EUR is home to 3.700 academics and professionals and almost 33.000 students from more than 140 countries. Everything we do, we do under the credo The Erasmian Way – Making Minds Matter. We’re global citizens, connecting, entrepreneurial, open-minded, and socially involved. These Erasmian Values function as our internal compass and create EUR’s distinctive and recognizable profile. From these values, with a broad perspective and with an eye for diversity, different backgrounds and opinions, our employees work closely together to solve societal challenges from the dynamic and cosmopolitan city of Rotterdam. Thanks to the high quality and positive societal impact of our research and education, EUR can compete with the top European universities. www.eur.nl.

SocialMovez

Online social networks such as Instagram, TikTok and Snapchat offer unprecedented technological possibilities to promote a healthy lifestyle among young people. For example, through influencers in their online network. With our research, we want to make online health campaigns more effective while safeguarding young people’s digital privacy.

SocialMovez investigates how health campaigns for young people can be spread more effectively through peer influencers in online social networks. We focus on promoting healthy behaviors such as physical activity or healthy eating habits. In our research we aim to identify peer influencers in young people’s online social networks and motivate them to engage in the health campaign.

We use innovative technologies in the field of data analysis to optimize online (health) campaigns without compromising the privacy of young people. We co-create our campaign messages and strategies with young people, parents, and health professionals.

MediaMovez

Young people grow up in a world where computers, tablets and smartphones are omnipresent. These digital media technologies offer important opportunities in the fields of entertainment, education, communication and cultural development. This being said, there are also risks involved that can affect their well-being and safety. Our research highlights the ways in which young people can take full advantage of the opportunities that digital media provides. Simultaneously, we emphasize how youth can cope with the potential risks of digital media use. Education to promote media literacy is indispensable for this.

MediaMovez helps in the development of effective media literacy education programs. Through research, we can contribute by developing education programs that help children interact with digital media in a more autonomous, safe and responsible way.

MediaMovez uses a media empowerment approach. This approach aims to strengthen the skills that young people need to use digital media independently and in a media-smart way.

Contact Movez

Linkedin

Website

movez@eur.nl